Relevance of average summation in top output

[Ralph Grothe]

One of our clients' DBAs asked me for an explanation regarding the strange avg line in top's display (see dump below)

I usually don't give too much care for such average values and am suspecious about percentages of all sort unless I'm not sure what baseline they relate to.

But can anyone help me out with an explanation?

Regards
Ralph

Load averages: 0.51, 0.54, 0.48
791 processes: 781 sleeping, 10 running
Cpu states:
CPU LOAD USER NICE SYS IDLE BLOCK SWAIT INTR SSYS
0 0.42 9.9% 0.0% 0.0% 90.1% 0.0% 0.0% 0.0% 0.0%
1 0.36 2.0% 0.0% 0.0% 98.0% 0.0% 0.0% 0.0% 0.0%
2 0.54 11.0% 0.0% 3.0% 86.0% 0.0% 0.0% 0.0% 0.0%
3 1.03 3.0% 0.0% 1.0% 96.0% 0.0% 0.0% 0.0% 0.0%
4 0.39 20.0% 0.0% 3.0% 77.0% 0.0% 0.0% 0.0% 0.0%
5 0.33 11.0% 0.0% 4.0% 85.0% 0.0% 0.0% 0.0% 0.0%
--- ---- ----- ----- ----- ----- ----- ----- ----- -----
avg 0.51 0.0% 0.0% 0.0% 100.0% 0.0% 0.0% 0.0% 0.0%

Memory: 2257364K (461896K) real, 2175324K (450676K) virtual, 2376792K free Page
# 1/99

[Paula J Frazer-Campbell]

Ralph

The average row in top is only the average of the column over the last polling period. Which does not tell you very much.

Top is not exactly the worlds best tool to look at what is happening on your servers.

GlancePlus will give far more accurate and detailed information.

A 60 trial version is on the CDs.


Paula.

[Steve Steel]

Hi


There is a document logged in Kmine which states that the avg line is badly calculated on multi processor systems.

The latest patches for top and pstat should be added to the system but are not yet guaranteed to cure it in all cases.

As indicated ,if you really want to see what is happening use glance.

Top is at best a first pass diagnostic tool .


Steve Steel.

For 11.00
The latest top(1) and pstat patch ( [PHCO_22686/PACHRDME/English] and [PHKL_22986/PACHRDME/English] )

[Ralph Grothe]

Paula, Steve,

thanks for guiding me towards glance, gpm.
In fact there is no need for me to use the evaluation trial version since we do have full licenses for most of our servers.
This includes also other perfomance tools such as MWA, PRM, WLM etc.
So I do appreciate glance, and do use it.
It was only because I had been asked for an explanation for top's seemingly erratic output and haven't been able to give one :-(
That's why I wanted to ask the experts here.

Steve, thank you for providing me some patches for top.
Unfortunately a kernel patch isn't an option right now (owe to reboot), but I may have a look at the commands' patch.